Mapping narrative reports of drug induced experiences: A data-driven dictionary-based approach.
Abstract
<p>Firsthand accounts of drug users’ experiences offer a rich resource to enhance our understanding of the effects of different substances. Advances in natural language processing (NLP) and machine learning (ML) techniques offer promising ways to interrogate these subjective reports at scale. The present study applies a broad-scope, data-driven, dictionary-based approach to a large corpus spanning 56 substances across multiple pharmacological classes. Using a pipeline combining random forest classification and k-means clustering, we identified 22-word categories most informative for differentiating substances and examined how substances grouped according to shared linguistic profiles. Three broad clusters emerged: a Hedonic-social cluster characterised by social, motivational, and affective language; a Pharmaceutical-instrumental cluster marked by biological, somatic, affective, and functional language; and a Psychedelic-like cluster distinguished by perceptual, spatial, and insight-related language. These groupings show areas of convergence with prior classifications alongside informative divergences, suggesting that word-category frequencies capture meaningful but coarse-grained regularities. Notably, clusters did not map neatly onto pharmacological classes alone but instead appear to reflect a combination of pharmacological features, phenomenological features, contextual and functional use patterns, and motivational orientations. While abstraction from broader semantic context introduces ambiguity and warrants cautious interpretation, the present findings demonstrate that data-driven word-category features can reveal structured patterns across a diverse range of substances. Such approaches show potential for continued refinement and offer a broad-stroke entry point into large-scale semantic mapping that can serve as a scaffold for progressively finer-grained, theory-informed, and multi-method investigations of drug induced experiences.</p>
Topics & conditions
Detected automatically by Psychedelic Commons from the title and abstract, not labels from the source. They may be incomplete.
Psychedelic Commons aggregates metadata from open sources and links back to the original. Rights remain with the authors and publishers. Something wrong with this record? Report a problem.